Tech Talk Interviews

You've been speaking about AI governance at this summit for four years now. How has the conversation evolved?

Alex: AI is not new, and AI governance is not new. Four years ago, we were talking extensively about machine learning models, bias, and predictions based on underlying data. But over the last couple of years, the excitement and energy around AI has greatly accelerated, first with large language models and especially over the last year with agents.

At IBM, we work with organizations around the world, and we understand what it takes to really get it right for the enterprise. That's what inspired us to conduct this study on the impacts of AI governance, looking at what organizations are doing well and what the opportunity cost is for not getting it right.

What did the study reveal about the current state of AI governance?

Alex: The findings are quite striking. Over half of the organizations we surveyed don't identify as having a robust framework for AI governance, despite the fact that they plan to put a major AI initiative into production within the next year. There's clearly a disconnect here.

Conversely, when we talk to organizations that do have a robust AI framework in place, they see major accelerants to their business. Organizations with strong governance frameworks saw a 27% increase in efficiency gains from AI use cases and a 34% increase in operating profit from those use cases. There's a real case to be made that governance helps you be more efficient, improves your bottom line, and acts as an overall accelerant to the whole process.

What specifically makes AI governance so valuable for businesses?

Alex: When we look at organizations doing this well, you can get quantifiable numbers as to why they're performing better. They see faster time to value, improvement in their regulatory risk profile, and elimination of bottlenecks and late stage problems that plague organizations and ultimately pull those use cases and all that work out from production.

What practical steps can companies take to implement effective AI governance?

Alex: One of the things I love about the study is it outlines several really practical things organizations can do. First, think about the data. Three quarters of the organizations we surveyed say that data management is a key barrier for them in their AI use cases. As it has been with analytical work or being more data driven, organizations really need to think about the data first and foremost.

Second, think about the end to end process. Organizations get excited about AI, but if you actually map out on paper from end to end, from the idea to actual execution, who are the people that need to be involved? What are the technologies? What are the handoffs? What are the workflows? Try to identify specific gaps in what you're currently able to do with just one high risk use case on paper. It's really quite helpful.

How should governance frameworks adapt as AI technology evolves?

The third recommendation is to move from a static AI governance policy where things are hardcoded into systems into something that's very dynamic. Think about how you constantly iterate and improve as the technology changes, as the world changes, and improve what your agents are able to do.

Another critical aspect is embedding governance into the workflows, not making it just part of a quarterly checklist where you say you did your review and check a box. How are you baking it into all that you do as you go forward?

Moving from static to dynamic is really framed in almost agile thinking. What you want to do is constantly evaluate how the use case has performed, what the issues have been, and how you constantly iterate and improve. At IBM, we've worked with organizations around the world and see a need for tooling to help do that. That's why we've built things like Watsonx Governance and Watsonx Orchestrate to enable that holistic governance and agent ops approach.

Can you explain what these tools do?

Alex: We've had Watsonx Governance in market for a couple of years now. It's really a holistic toolkit to enable AI governance, everything from lifecycle management of thinking about what the use case is, who approves it, what the business workflow is for moving from idea to production, as well as having monitoring capability for risk management and a regulatory compliance aspect. As there are different regulations around the world, it helps you understand the implications of those use cases.

The second thing we've invested heavily in this year is Watsonx Orchestrate. That's IBM's agentic set of capabilities. A big focus this year has been agent ops, which is what I was talking about before. It's not necessarily a product but more of a concept of thinking about the full lifecycle management of the agent development lifecycle for whatever your use cases are.

What does agent ops enable?

Alex: We've focused on enabling organizations to have observability, tracing, and optimizations, really treating this as a virtuous cycle so they can do even better with what they're doing from agents, regardless of what agents they use. It doesn't have to all be IBM. It can be any provider and any deployment.

IBM in 2020 really refocused its strategy on hybrid cloud and AI. We work with clients around the world solving some of their most difficult challenges, tapping into these two mega trends where we see major opportunities to be efficient, help your bottom line, and really outpace your competition.

We don't live in a world where anybody is going to have just one cloud provider, one solution, one SaaS based application, or one on premises infrastructure. They're going to have a myriad of things, a hybrid cloud estate, and that's where we have a lot of focus.

What are the key enterprise considerations for AI that you're helping clients navigate?

Alex: If we had this conversation maybe a year and a half ago, we would have talked just about large language models. But now we're really thinking about the enterprise considerations of AI, whether that's cost optimization, governance, transparency, and those types of things. There's tremendous excitement about agents, but working through these considerations with clients is where we see the real value.

We just launched this new AI governance study through IBM's Institute for Business Value, the IBV. It provides detailed insights into how organizations can build robust governance frameworks and the measurable business impact of getting it right.

Conclusion:

Wylie's insights make clear that AI governance isn't just a compliance checkbox but a strategic advantage that directly impacts efficiency, profitability, and competitive positioning. As AI capabilities continue to advance, the organizations that build dynamic, embedded governance frameworks today will be the ones that succeed tomorrow.